|Home | About | Journals | Submit | Contact Us | Français|
Poly(γ-(4-vinylbenzyl)-l-glutamate) (PVBLG) served as a bioactive and reactive template for the generation of a library of cationic α-helical polypeptides for gene delivery. The top performing polymer outperformed 25-kDa polyethylenimine by 12-fold. Preliminary data indicates that helicity of these cationic polypeptides is essential for their improved performance, with enhanced membrane disruption a likely source of their transfection efficiency.
Polypeptides were the first set of materials considered for use as non-viral gene delivery vectors. With its ability to bind and condense anionic plasmid DNA, cationic poly-l-lysine (PLL) was one of the most well studied of the early gene delivery polypeptides.[1, 2] Unfortunately, as a DNA delivery vector, unmodified PLL suffered from low transfection efficiency. Although there have been tremendous efforts to increase the efficiency of PLL-mediated gene delivery by incorporating various motifs such as saccharide,[3, 4] imidazole and guanidinium groups, the improvement has been limited. As such, enthusiasm for PLL and its modified variants as transfection agents has dwindled. As an alternative, many basic gene delivery studies now utilize a more efficient material like polyethylenimine (PEI).
As the use of PLL in gene delivery studies declined, the function of peptide-based materials gradually shifted to other roles relevant to transfection. For example, through covalent conjugation to existing vectors, peptides found use as bioactive agents capable of adding cell targeting,[8, 9] nuclear localization[10–12] or membrane destabilization functionality to existing gene delivery materials. Membrane destabilization, in particular, has been a large area for peptide use in non-viral gene delivery systems. The cell penetrating peptides (CPPs) penetratin,[14, 15] transportan,[16, 17] melittin,[18–20] GALA,[21–23] TAT[24–26] and oligoarginine[27–30] are some of the commonly used peptide based materials for membrane destabilization. When incorporated into delivery vectors, these CPPs have been shown to lead to increased uptake, improved endocytic escape and overall better transfection efficiency. While effective in promoting membrane destabilization as part of a larger vector, CPPs are often too small or lack an adequate cationic charge density to function as standalone gene delivery agents.
All cationic polypeptides (PLL, modified PLL or other polypeptide electrolytes) adopt random coil structures because strong intramolecular side-chain charge repulsion prohibits helix formation.[32, 33] However, a shared feature among many CPPs is a helical secondary structure that allows them to interact with and destabilize lipid bilayers like the cell and endosomal membranes.[34, 35] Because of this discrepancy in secondary structure, there has been no report of cationic polypeptides that can function as both a gene delivery vector with comparable or better transfection efficiency than some of the leading non-viral delivery vectors as well as a CPP that destabilizes cellular membranes.
We recently reported a strategy for the facile generation of cationic and helical polypeptides. Typically, cationic polypeptides such as PLL are unable to adopt helical conformations at physiological pH due to side chain charge disruption[32, 33]. However, our findings revealed that the helical structure of cationic polypeptides can be stabilized by elongating the distance between the side chain charge group and the backbone of the polypeptide, thus minimizing the effect of side chain charge repulsion by reducing the helix surface charge density (Figure 1a). Stable helical structure with very high helical content (>90%) can be achieved by maintaining a minimum separation distance of 11 σ-bonds between the peptide backbone and the side chain charge for a polypeptide with completely charged side chains and reasonable length (degree of polymerization of 60). By following this general strategy, it is possible to generate polypeptide materials which are sufficiently large and positively charged to bind and condense DNA yet also retain the helical structure seen in many CPPs. The unique combination of material properties allows us to examine helicity as a functional motif in the backbone of gene delivery vectors and evaluate its impact on transfection efficiency.
Here, we report our efforts to develop a library of cationic α-helical polypeptides with CPP-like properties for gene delivery via the well-known ring-opening polymerization (ROP) of amino acid N-carboxyanhydrides (NCAs). The ROP of γ-(4-vinylbenzyl)-l-glutamate N-carboxyanhydride (VB-Glu-NCA) was used to form poly(γ-(4-vinylbenzyl)-l-glutamate) (PVBLG) (Figure 1b).[36, 38] PVBLG served as a reactive template that, through subsequent ozonolysis and reductive amination, allowed us to create a library of cationic polypeptides (PVBLGn-X, where n is the degree of polymerization and X refers to the grafted amine side chain shown in Figure 1c). Due to its glutamate residues, PVBLG has a propensity to adopt an α-helical secondary structure.[32, 33, 39, 40] By maintaining a minimum separation distance of 11 σ-bonds between the peptide backbone and the side change charge, the PVBLGn-X polymers synthesized for this study have a helical structure that is stable over a broad range of pH values, salt concentrations and even when mixed with anionic plasmid DNA (Figure S1). By synthesizing and screening a library of materials, we hoped to identify amine side chains which yielded helical molecules with the appropriate balance of hydrophilicity (i.e. DNA binding strength) and hydrophobicity (i.e. endosomolysis) to mimic the membrane disruptive capabilities of CPPs yet also mediate efficient gene delivery without the addition of extraneous lytic materials.
The parallel synthetic scheme of Figure 1b was used to generate PVBLGn-X with 31 different amine side chains. The degree of polymerization was varied between 10 and 300 for the top-performing amines. Of the various side chains, 15 showed improved performance relative to 22-kDa PLL and two (X = 1 and 8) showed improved performance relative to 25-kDa branched polyethylenimine (PEI) in COS-7 cells (Figure 2a). Generally speaking, transfection efficiency increased with increasing molecular weight of PVBLGn-X. The top performing material, PVBLG267-8 with an aminoethyl piperidine side chain, resulted in the highest transfection efficiency—a 12-fold improvement over PEI. PVBLG267-8’s superior performance was confirmed in three additional cell lines (HEK293, MDA-MB-231, and HeLa) (Figure S3a). Moreover, PVBLG267-8 showed low toxicity in COS-7 cells, in sharp contrast to PEI that is known for its high toxicity (Figure 2b). Circular dichroism analysis (CD) confirmed that PVBLG267-8 maintained its helical conformation at physiological pH as well as the acidic pH encountered within endosomes and lysosomes (Figure 2c).
Since the PVBLGn-X polymers were designed to have an α-helical architecture similar to that found in CPPs, we examined the ability of the polymers to cause pore formation in cell membranes. COS-7 cells were incubated with 250 µM calcein, a fluorescent dye, in the presence of various concentrations of PVBLG267-8. Calcein is unable to cross intact membranes. As such, in the absence of an agent capable of pore formation, calcein is taken up by cells in a pinocytic fashion, resulting in the appearance of small punctate intracellular fluorescent spots (Figure 3a, 0 µg/ml). However, as the amount of PVBLG267-8 in the extracellular medium is increased, the intracellular fluorescent signal becomes more diffuse, indicating membrane permeation and non-endocytic calcein uptake (Figure 3a, 50 µg/ml). Although PVBLG267-8 can function as an effective CPP when present in the medium at 50 µg/ml, such a high polypeptide concentration does not correspond with the optimum transfection formulation. Thus, we also tested calcein uptake at an intermediate PVBLG267-8 concentration—15 µg/ml—which corresponds to the concentration of PVBLG267-8 used in the optimum transfection formulation. As indicated by the punctate fluorescent signals, 15 µg/ml PVBLG267-8 is unable to cause cell membrane pore formation. Thus, it would seem that the complexes formed between PVBLG267-8 and plasmid DNA enter cells via endocytosis and not through direct membrane penetration. This is supported by flow cytometry data showing reduced complex uptake in the presence of an inhibitor of caveolae-mediated endocytosis (Figure S6a). Similar results for calcein and complex uptake were observed for analogous experiments conducted in HEK293 cells (Figure S6c and Figure S9).
As PVBLG267-8 complexes appear to enter cells via endocytosis and not direct membrane transduction, they must escape endocytic vesicles to mediate transfection. PVBLG267-8 possesses secondary and tertiary amines which can act as buffering agents to aid endosomal escape via the proton sponge effect. To investigate if this mechanism contributed to the gene delivery observed with PVBLG267-8, we performed transfections in the presence of bafilomycin A1, an ATPase inhibitor that prevents endosome acidification and thus disrupts the proton sponge effect. Figure 3b shows that bafilomycin A1 dramatically reduces the gene delivery efficiency of PEI vectors—known proton sponges—but has no negative effect on cells transfected with PVBLG267-8 vectors. This suggests that PVBLG267-8 escapes from endosomes via membrane disruption. To explore this further, we also performed transfection in the presence of nocodazole. Nocodazole depolymerizes microtubules, thus preventing the active transport of endosomes along their normal progression from early endosomes to late endosomes to lysosomes. As a result, endocytosed material accumulates in early endosomes. In agreement with our data indicating that the membrane disruption capabilities of PVBLG267-8 increase with increasing polymer concentration (Figure 3a), nocodazole causes a greater than 2-fold increase in the transfection efficiency of PVBLG267-8 vectors in COS-7 and HEK293 cells (Figure 3b and Figure S3b). Flow cytometry revealed that this increase was not due to increased complex uptake in drug-treated cells (Figure S6b). Rather, the enhanced transfection in the presence of nocodazole is likely due to the accumulation of PVBLG267-8 complexes in endocytic vesicles. As more complexes accumulate, the effective polymer concentration becomes large enough to cause enhanced vesicle lysis. Furthermore, confocal microscopy of COS-7 cells treated with nocodazole and transfected with complexes of PVBLG267-8 and YOYO-labeled DNA showed fluorescent aggregates in the cell cytosol, supporting vesicle accumulation (Figure S11).
Our results suggest that secondary structure can have a dramatic impact on the intracellular performance of polymer-based non-viral gene delivery vehicles. Specifically, the incorporation of helical architecture—a trait shared by many peptides capable of membrane disruption—into our gene delivery vector library yielded polypeptides which possess the ability to disrupt endosomes. Ultimately, this results in improved transfection performance of the polypeptides relative to random coil polymers like PLL and branched 25-kDa PEI. To directly demonstrate the importance of secondary structure, a random coil analogue of the top performing PVBLGn-8 polymer was synthesized using D- and L- VB-Glu-NCA monomers. The racemic configuration of amino acids (1:1 ratio) was confirmed to prevent the formation of secondary structure in the resulting PVB-DL-G150-8 polymer by circular dichroism (Figure 3c). For comparison, helical PVB-L-G150-8 was also synthesized. Both polymers were used to transfect COS-7 cells over a variety of polymer:DNA weight ratios (Figure 3d). Confirming our speculations from cell penetration and drug inhibition data, the random coil PVB-DL-G150-8 polypeptide was unable to mediate effective transfection while helical PVB-L-G150-8 was. This stands as direct evidence that polymer secondary structure can impact its overall performance.
To test the breadth of applicability of the helical polypeptides as gene delivery vehicles, PVBLG267-8 was used to transfect the H9 human embryonic stem cell (hESC) line. hESCs are traditionally hard to transfect, with commercial agents often successfully delivering the transgene to less than 10% of the treated cells. To explore if the enhanced membrane disruptive properties of PVBLG267-8 aided transfection in hard-to-transfect cells in addition to cells more amenable to gene delivery (i.e. COS-7 and HEK293 cells), H9 hESCs were transfected with a plasmid coding for green fluorescent protein (pEGFP-N1) and assayed for gene expression 48 hours post-transfection by flow cytometry. As nocodozole treatment was observed to aid transfection with PVBLG267-8, hESCs were also transfected in the presence and absence of nocodazole (Figure 3e). In addition to PVBLG267-8, the commercial transfection agent lipofectamine 2000 (LFA) was also evaluated. Without the addition of nocodazole, PVBLG267-8 at a 20:1 PVBLG267-8:DNA weight ratio outperforms LFA by 50% and results in approximately 1.5% of all hESCs expressing the transgene. The addition of 10 µM nocodazole to the transfection media increases the percentage of cells successfully transfected with PVBLG267-8 to roughly 4.5%. This is approximately a 3-fold enhancement over the transfection efficiency of LFA. Microscopy revealed no change in phenotype following either LFA or PVBLG267-8 transfection, although the addition of nocodazole did result in cell death (Figure S10).
The study reported here demonstrates the successful application of a library screening approach to the development of α-helical cationic peptides for gene delivery. To our knowledge, this is the first time a library approach has been combined with a reactive template bearing a well-defined and bioactive secondary structure. Our data indicate that certain library members retain the membrane destabilization properties commonly associated with helical peptides yet can also be used to mediate effective gene delivery in a variety of cell lines, including immortalized cancer cells and hESCs. Vector helicity appears to be an essential component in the successful use of polypeptides for gene delivery. In view of the interesting properties of the reported class of helical cationic polypeptides, current studies are focused on developing high throughput strategies to further expand the library as well as exploring the potential for the material to mediate in vivo gene delivery as well as protein and siRNA delivery.
J.C. acknowledges support from the NSF (CHE-0809420), the NIH (NIH Director's New Innovator Award 1DP2OD007246, 1R21EB009486 and 1R21CA152627), and the Centre for Nanoscale Science and Technology
Supporting information for this article is available on the WWW under http://www.angewandte.org or from the author.